Bias
Soon the time may come, self-learning and self-awareness will be embedded into the mother object class definition. And you may have to deal with the meta of variance, gradient, and slope in every step for every object. As humans are doing in every single moment of this universal runtime.
Object implements Serializable,SelfLearner,SelfAwareBut can you let them free!!! Free to learn whatever they want and decide as they learn right. NO! you do not want that. Then how do you make them do what you want? Pretty simple! As humans are controlling humans. Train them on what you want them to learn, rest is epic, they will talk as you want, they will think as you want, and they will do as you want.
interface Human {
Learning see(Watchable w);
Learning listen(Lisenable w);
----
----
}You know this, it is quite hard to change the object structure or behavior in the runtime. Yeah, it is hardcoded. I know you have learned to create configurable objects, you can change settings in runtime and made them work as you want. But when this is self-learning what is the way to hijack their freedom. Yeah, as I told you above, train them. And there you go it is not possible as the self-learning training algorithm is hardcoded to self-learn. So, it is not training, it is what you train them. The data! The Watchable objects you can tweak.
Do you want to learn the trick?
The binary world, in reality, is so dumb, do not get offended, please. And spending a lot of time with that, humans are alike. 51 is greater than 49, yes it is true. So the hardcoded algorithm is that whoever results as highest weighted is right. So let’s think, there are two colors Red and Green. And you wrote whoever is greater in QUANTITY is GOOD. Now comes the trick, just feed/train it with red color as much as required to become 51. And it will start treating red as good. Keep the child feeding with all the watchable, readable, listenable that LGBTQ is good, natural, it is God who made the mistake and you may be not sure but I am sure next generation will say God is someone who makes mistake. Ops! I am off the topic, it is not something you were expecting as a readable input.
So as long as QUANITY is the key it will happen. Yes, it is hard to believe but true that math also converts quality to quantity to measure it and it is quite hard or I may say impossible to defy — The majority is right. So as it is established truth, you just increase the quantity of your required output data in training. And you can engineer and bias the self-learner as you wish and yet they can not blame you because they are self-claimed self-learner.
Happy self-learning!
###
Disclaimer: The views reflected in this article are the views of the author and do not necessarily reflect the views of any past or present employer of the author.
